AI Consultant — LLM Architecture - 2-month mission at 40% extendable

Mission Context

The bank has a language model (LLM) already operational on its internal infrastructure. It is looking for an AI consultant to evaluate the existing setup, define the target architecture, and implement the developments necessary for its securing and its availability within the bank.

The mission combines consulting, architecture, and implementation, in collaboration with internal teams and the offshore IT team based in Asia.

Responsibilities

• Conduct an assessment of the model, the on-premise infrastructure, and the current access and operation methods.
• Identify risks, technical gaps, and security needs.
• Define an architecture adapted to the requirements of the banking information system: integration, access control, data protection, traceability, and supervision.
• Develop a prioritized roadmap for making the LLM available to internal users and applications.
• Implement the selected developments and support their integration into the IS.
• Coordinate work with the offshore IT team in Asia and the bank's infrastructure and security contacts.
• Document the solution and ensure the transfer of knowledge to the teams responsible for its operation.

Profile sought

• Concrete experience in the deployment and operation of AI solutions, notably LLMs, on on-premise infrastructures.
• Skills in architecture and integration of AI solutions into existing information systems.
• Solid knowledge of IS security: authentication, authorizations, network security, data protection, and logging.
• Ability to intervene in both consulting, architectural choices, and technical implementation.
• Experience collaborating with international or offshore IT teams.
• Professional English allowing exchanges with the team based in Asia.
• Experience in a banking environment or in a sector subject to high security requirements is an asset.

Expected Deliverables

• Technical and security assessment of the existing solution.
• Architectural recommendations and prioritized action plan.
• Implementation of the agreed developments to secure and make the model accessible internally.
• Technical documentation and operating instructions.

j4id10439554jm j4it1042jm j4iy26jm